On 12/14/2017 03:38 AM, Paolo Bonzini wrote: > On 14/12/2017 01:59, John Snow wrote: >> qemu_coroutine_yield(); >> + job->last_yield_ns = qemu_clock_get_ns(QEMU_CLOCK_REALTIME); > > This is not the time the job has yielded control, but the time the job > has gotten it back. Is it intended? > > Thanks, > > Paolo >
Yes, since that matches how mirror recorded last_sleep_ns, except for where it records the 0ns sleep, then it records it beforehand (close enough.) I intended for this valuable to count how long we haven't yielded, basically. Thought being that any time we are yielded, we're being cooperative. This is the selfishness counter :)